How to create custom sapi voice for tts

Question

I am working on a project that I need to create a custom voice engine for my application. I saw something like TTS Builder, but is there anyone who understands how applications like TTS Builder itself are developed? What is the problem with SAPI engines? How do they work? How can you build it? Can I develop my own algorithm? I would rather do it in C # if possible

Answer 1

From what I see, it looks like TTS Builder accepts existing voices and allows you to change minor parameters to make a slightly different voice. But creating a voice with a different accent or pronunciation is, I think, more difficult.

From AT&T Research :

Creating high-quality voices requires a good voice, sound insulation, professional audio equipment, hours of written material with careful coverage of combinations of phonemes in the language, as well as time and experience, in order to turn these recordings into a worthy synthetic voice. Due to the costs involved, custom voice assemblies are usually performed for corporations that want to computerize an actor’s existing voice, for example, to continue branding.
...
Building a transformational model may require much less material than creating a TTS voice from scratch.